Discourse actually has some awesome bot or otherwise “suspicious account” detection and alerting. Most of the time their comments get caught before they are even displayed.
No, it takes two reports from active community members to autoremove posts.
As max says, discourse has some pretty nice tooling for this such as trust levels, sock puppet detection and botlike activity. If anything it is too sensitive.
While those accounts on Twitter might be just reposting and reply guys they aren’t necessarily bots. We’re back to the old sybil problem where the only thing that matters in these discussions is stake, followed potentially by reputation.